According to Kumārila, much can be sense-perceived. It goes without saying that sensible qualities can be sense-perceived, but Kumārila thinks that we can also sense-perceive the substance behind the sense-qualities (that is, the substrate of the sense-qualities). He also thinks that we can sense-perceive the universal inhering in the particular. Thus, when we look at a brownish cow, we are sense-perceiving its colour, the substance-cow and the universal-cow.
However, this rather generous account of perception comes with some serious and specific boundaries. Perception, to begin with, is only about the present. It cannot grasp the past nor the future. There is no yogic super-sensuous perception that would be able to grasp such features of reality.
Kumārila also denies that cognitions are self-aware (this self-awareness, or svasaṃvedana is considered to be a form of perception by Buddhist epistemologists) and rather claims that we become aware that a cognition has taken place only retrospectively, through arthāpatti. Thus, besides denying self-awareness he also denies the Naiyāyika anuvyavasāya or `apprehension of a previous cognition’, through which one becomes aware of a previous mental event. Why is anuvyavasāya not acceptable for Kumārila? Presumably because it is about something no longer present (this might be the main reason for his general denial of mānasapratyakṣa if it is about prior thoughts).
The only seeming exception is ahampratyaya `cognition of ourselves qua-I’, which grasps something other than an `external’ object. Kumārila still thinks that the I is not a construct, but something objectively real, but he claims that in that specific case we have direct access to it. How exactly is still under investigation (see my previous posts on the matter), but my current understanding is that ahampratyaya grasps the I-as-knower while it is knowing something. It cannot grasp a previous I, otherwise it would violate the boundaries of sense-perception discussed above.
Thus, as much as Kumārila is generous with regard to regular sense-perception, he is strict in denying any sort of perception beyond it.
Kumārila’s work appears against the background of classical debates concerning universals, substances and particulars, involving Jains, Grammarians and Buddhists. He is well known for admitting perceptions intermediate between universals and particulars.
In Radhakrishnan’s History this impact is noted also in relation to language, allowing levels intermediate between the subtle and coarse or manifest. Now Hume Treatise 3.3.4) contrasted his view of moral sense with the tradition of Grammarians, who merely noted terms of moral valuation. For Hume, pleasure and pain are universals of sense, but moral qualities are more varied, as involving something of the agent and the purpose or object.
Could it be that Kumārila understood perception of self in a similar way, as the sense of ‘how one is doing’ with the opportunities and challenges of life? Looking again to Ryle’s distinction, we then have the sense of self in the mode of knowing how, not giving self as an explicit or distinct object of knowledge.
Hi Orwin, thanks for your question. I am not sure I am following your train of thoughts, e.g. when you write “Now Hume Treatise 3.3.4) contrasted his view of moral sense with the tradition of Grammarians, who merely noted terms of moral valuation.” Which Grammarians is Hume attacking? Or is something missing in this sentence?
Last, what do you mean by “Kumārila understood perception of self in a similar way, as the sense of ‘how one is doing’”?
Thanks and best.
Some thoughts: I would suspect that manas is involved here, since perception of pleasure, pain, etc., count as “perception” according to Kumārila. And the manas and ātman are connected/saṃyukta, such that one can directly perceive one’s internal mental states/cognitions. (discussed in the perception section).
And given the way that Kumārila talks about the relationship between the cognizer and the cognition (there is no third category of inherence that ties them, and the cognitions are states or conditions of the cognizer), if one is able to have direct experience of a cognition, direct experience of the cognizer is not far off. Showing that there is an underlying, substantial essence, not a mere causal series of cognitions is the difficulty, and he gives many (Nyāya-influenced) arguments for this against his Buddhist opponent in the ātmavāda. However, beyond inference, it seems that if pratyabhijñā (in vv. 117ff) of the self is a kind of mental perception, then one can perceive not only the self but the self as an enduring thing, like when we reflect, I know now what I knew then. (And since, as Kumārila argues against his Buddhist opponent, the ahampratyaya is not mistaken, our recognition counts as a genuine pramāṇa, cp. śabdanityatā on pratyabhijñā in the context of words, 372 especially.)
Knowing is a mental condition of the self with which we are directly acquainted, perceptually, consistent with what you suggest at the end of the post. And we recognize the content as the same, which is still a sense perception.
Last thought is whether *that* the self is an enduring thing is inferred or known by recognition. I think it’s possible that Kumārila might say we know the self as it is through recognition, which is grounded in (mental) perception. But knowing that the self is a single enduring thing requires an inference from recognition, just as the nityatā of śabda is not conclusively known by recognizing the same recurring word type in different tokens. We must do more to show this. We must make an inference. But we do recognize the same word type in each utterance.
Many thanks for sharing your thoughts, Malcolm!
Yes, I agree about manas-pratyakṣa re. the self and about the role of pratyabhijñā (which, any way, needs a perceptual component, so it cannot work independently of it).
However, I think that manas cannot grasp cognitions (I don’t know whether you meant that, but it seems to be implied by “one can directly perceive one’s internal mental states/*cognitions*”, by “if one is able to have direct experience of a cognition, direct experience of the cognizer is not far off” and by “Knowing is a mental condition of the self with which we are directly acquainted, perceptually”), because of Kumārila’s insistence on cognitions not being directly accessible, because they are opaque. Instead, we know that we have known something because of arthāpatti (as you know very well!).
Now, the situation becomes complicated because the self is known perceptually-qua-knower and yet knowledge is not known perceptually but only known via arthāpatti. Sucarita even says that the self is known through its dharma being cognition (see here: https://elisafreschi.com/2026/02/26/again-on-ahampratyaya-in-kumarila-using-watson-2010-and-2020/). Thus, there seems to be a direct grasp of the self while it knows, as well as a direct grasp of the objects being known, but not of the knowledge activity itself. I am still thinking about it.
I’m thinking about verse 83 of the pratyakṣasūtra:
manasas tv indriyatvena pratyakṣā dhīḥ sukhādiṣu
manasā samprayukto hi tānyātmā pratipadyate
*That* we have had a cognition is not directly perceptible and must be known by positing the existence of a cognition that has content. But we are aware of cognitions immediately in that we are aware of their content, their objects. And for the self, it is even more immediate, in that we are immediately, introspectively aware of pleasure, pain, etc., through the manas. We certainly aren’t positing that we are having pleasure.
In fact, in his commentary on this verse, Sucarita has an opponent raise your worry, arguing that on this view, Kumārila would accept svasaṃvitti. He responds:
ayam abhiprāyaḥ. na sukhādisaṃvedanāny api vijñānāntaravidhām ativartante. tāny apy antaḥkaraṇena manasendriyeṇa saṃyuktāny ātmaguṇāntarbhūtāny evātmanā prātapadyante (page 239 of the TSS edition, volume 1).
This is the idea: experiences of pleasure, pain, etc. do not outstrip other kinds of cognitions. These are also connected with the internal mental sense faculty, being just internal properties of the self, they are understood by the self.
He goes on to say more about how these cognitions are inherent in the self and how we are still able to have distinctions among the knower, the the means of knowing, the object of knowing, and the cognition (act of knowing). I think this discussion is relevant for what’s happening in the ātmavāda. And it coheres with what Watson’s paper says about Sucarita.
But again, I see no problem with direct access to pleasure, pain, and the contents of cognitions. The issue is the cognition grasping itself directly as both instrument and object. And that’s not happening here.
Sorry, just one more point:
“Kumārila’s insistence on cognitions not being directly accessible, because they are opaque. Instead, we know that we have known something because of arthāpatti.”
1. Maybe I’m misunderstanding, but where does Kumārila say cognitions are opaque? (I guess this depends on how you intend “opaqe.”) As I understand him, the intentional object of a cognition is what is presented to us in the act of knowing (experiencing, doubting, etc). Cognitions are transparent in that sense. But they are not reflexive, in that a cognition does not cognize itself, so that the cognition’s content/object cannot be the self-same act of cognizing.
2. Knowing *that one knows* (KK) is not the same as knowing (K). In an act of knowing perceptually that p, *that p* is directly available to me by perception. But by arthāpatti, perhaps I can know later *that I know* that p. (Always subject to svataḥprāmāṇya, of course.)
Consider an act of perception: The pot is blue.
T1: Perception gives me the content of a cognition, “The pot is blue,” (pot, blueness, in relationship)
T2: Arthāpatti can give me, “I know that the pot is blue,” (self, knowledge, in relationship to the pot, blueness, in relationship)
These are temporally separated. Applied to the self:
T1: Introspective perception: “I am happy,” (self, happiness, in relationship)
T2: Arthāpatti: “I know that I am happy.” (self, knowledge, in relationship to the self, happiness, in relationship)
The content/object of “I” is the self! (Hence the discussion of the ahampratyaya in the ATV.) And so part of the content known at T1 is the self and its happiness, just as part of the content known in the bāhyārtha case is a pot and blueness. Sure, we can say more about the nature of that self (enduring, substantial, etc.), but the self is part of what the manas gives us.
So, on how I’m understanding things, I think it’s important to distinguish between:
– Kp vs KKp
– knowing that I know there is a self vs knowing *that* there is a self vs knowing the self as part of a cognition
Hi Malcolm,
my apologies for the late reply, I had not received a notification of your comments.
—Yes, with “opaque” I just meant that cognitions disclose their content, but not themselves. We know that we knew the book we looked at because of arthāpatti.
—Through manas, I can perceptually apprehend sukhādi, I am not sure that I am apprehending the ātman-qua-endowed-with-sukha (see immediately below).
—Could it be that we are disagreeing about “manasā samprayukto hi tāny ātmā pratipadyate”? I take it to mean that the ātman, connected to the manas, grasps sukhādi. The ātman is not the object of the same cognition. At least not in this verse. And in the ātmavāda, when Kumārila goes back to the issue, he says that we grasp the ātman qua knower (not qua sukhādivat).
—To be honest, if we were to grasp the ātman-as-I through our perceptions of sukhādi, that would solve many of my doubts. But it does not seem to be what Kumārila says in ātmavāda, where he thinks that the ahampratyaya is about the ātman qua knower. What do you think?
Elisa, there’s nothing missing in the Hume quote: it’s a passing remark, and he died not return to the theme. But he is assuming the reader will know what he means, that the background so sketched was familiar. In that vein, one hears later of the historical-grammatical method in philology, as the old way everyone followed before higher criticism developed.
Priscian was influential in Rome in the era of the Indian Grammarians, but his lone impact doesn’t explain Hume’s retrospect. Noting that Brahmagupta’s treatise, with his theory of gravity, was translated into Latin in the eleventh century, I rather think the Indian philosophy of language had a broad influence. The Greek tradition offered only Plato’s Cratylus, which is very ancient in inspiration, more comparable to Vedic lore.
Back with your problem, Ryle placed knowing how prior to knowing that, so it falls in as the immediate, unreflective experience. His retrospect seems consistent Kumārila, but I’m not in a position to confirm that. My considered thought is different: that finding yourself is a moral saga, a matter of trust and consent. It was Amod’s puzzle about expressive individualism that got me thinking about this, so it maybe more apt here to later commentary and controversy. Like Malcolm’s final point involving knowing *that.
What ought to be kept a hold of in this discussion is that the mind is by nature non-conscious. The whole ‘field’ (B.G.13/the Field and the Knower of the Field) is a plane of physical interaction that purely by virtue of being/existing is pervaded by consciousness i.e. sat chit. ‘He shining everything shines after him’ . Self awareness is a natural result. Its location in the Body/Mind/Intellect (Ahamkara) comes as a result of the natural gravitation towards the sensing pole of the interaction. This is the distorted image of the self that the anatman doctrine controverts.